Providing Decision Support Anytime and Anywhere: MET System Experience

Wojtek MichalowskiUniversity of Ottawa, Canada

Roman Slowinski, Szymon WilkPoznan University of Technology, Poland

2

Outline

Decision support anytime and anywhere

Computer in a pocket

Classical DSS design

New architecture for anytime and anywhere DSS

Mobile Emergency Triage

Discussion

3

Providing Decision Support Anytime and Anywhere

Decision support is needed when a decision problem arises

This calls for use of computing devices that are more handy than desktop computers and that can work offline

Decision support is required for complex problems with sophisticated solution strategies (e.g. triage support and clinical caremap)

This calls for system’s design that allows for versatility and easy adjustment to problem on hand

Support anytime and anywhere – available on demand for a class of decision issues of diversified complexity

4

Computer in a Pocket

Mobile devices with “lean” computing capabilities(notebooks, tablets, handhelds, cell phones)

Issues Diversity of platformsDiversity of decision making settings (the same system may be used in doctor’s office and an ambulance)

SolutionVersatile decision support system that can be accessed on different mobile platforms and can support a range of decision issues

5

Classical DSS Design

Sprague (1980)

Model subsystem

Database subsystem

Interfacesubsystem

6

System’s Architecture for Classical DSS

Stand-aloneFunctionality confined to a single system

Client-serverFunctionality distributed between powerful server and "thin"client(s)

Client acts as a terminal hosting interface subsystemServer manages dialog with a user and hosts database and model subsystems

Typical example: web-based DSS

7

Issues with Classical DSS Architecture

Requires fairly “strong” computing platform But stand alone system is not feasible for many decision situations

Requires strong and stable connectionBut "thin" client – server system is not feasible for many decision environments

Requires uniform platform specifications for an interfaceBut hard-coding interface into a DSS is not feasible for many cross-platform implementations

8

Postulates

If computing platform is not powerful enough, a DSS should be “decomposed” into appropriate executable components

If a connection is not stable or constant, a "thin" client still should be able to operate and provide support

If computing platforms are heterogeneous, an interface should be adaptable to changing requirements

9

Our Solution

New architecture: extended client-server paradigmClient performs some tasks of a server while there is no connectivity

Model-based interface design

Interface model

Interfaceagent Physical interface

Platform model

10

New Architecture – Server

Problem subsystem

Interfacemodel

Decisionmodel

Solvermodel

Database subsystem

Solversubsystem

Synchronized witha client database

Builder IntegratorFrom a

clientTo a client

Interfacesubsystem

Platform subsystem

11

New Architecture – Client

Problem subsystem

Jr.

Solversubsystem

Jr.

Interfacesubsystem

Jr.

Database subsystem

Jr.

Executor

Interface

Model

SolverTo a serverFrom a

server

Synchronized with a server database

12

MET – Mobile Emergency Triage

Clinical support system to be used by the Emergency Department medical personnel for triage of patients with acute pain condition

Developed as a flexible DSS that can be accessed anywhere and anytime and used to support medical decision making with regards to a heterogeneous set of clinical decision problems (abdominal pain, scrotal pain, hip pain, etc)

Designed according to an extended client – server architecture. Client component can run on a desktop computer, a tablet computer, a handheld, and a mobile phone

13

MET Server

Problem subsystem

Abdominalpain

Scrotal pain

Hip pain

Solver subsystem

Heuristicclassifier Patients’

database

Synchronized witha client database

Builder Integrator

Platform subsystem

Cell phone

Handheld

Desktop

Interfacesubsystem

Pictogrameditor

List editor

Numeric editor

14

MET Client – Cell Phone

Problem subsystem Jr.

Solversubsystem Jr.

Clinical decision rules

Heursitic classifierInterface

subsystem Jr.Pictogram editor…

Database subsystem

Jr.

Executor (Symbian OS)

15

MET Client – Handheld

Pictogram editor…

Clinical decision rules

Heursitic classifier

Problem subsystem Jr.

Solversubsystem Jr.

Interfacesubsystem Jr.

Database subsystem

Jr.

Executor (PalmOS)

16

MET Client – Desktop

Problem subsystem Jr.

Solversubsystem Jr.

Clinical decision rules

Heursitic classifierInterface

subsystem Jr.Pictogram editor…

Database subsystem

Jr.

Executor (Mac OS X)

17

Discussion

Possibility to develop a versatile DSS (clinical system for complete patient management from triage to discharge)

Flexibility in receiving decision support (irrespective of a decision situation and a decision environment)

Ability to develop new theoretical model of the DSS design (to support wide range of clinical problems of varying complexity)

Ability to deliver cross-platform DSS implementation (system "learns" about a platform and performs self-adjustment to its specifications)

18

Acknowledgements

Rhonda Correll, Research Institute, CHEOKen Farion, Division of Emergency Medicine, CHEOGreg Forestell, Information Services, CHEOJoanne Ross, Information Services, CHEOJohn Pike, Division of Urology, CHEOSteven Rubin, Division of Surgery, CHEO

Mathieu Chiasson, MET Research TeamNataliya Milman, MET Research TeamBernard Plouffe, MET Research Team

AppForge MobileVBiAnywhere Studio